
#Disable warnings for clean presentation.
import warnings
warnings.filterwarnings("ignore")


import pandas as pd
import math
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
import datetime as dt
import missingno as msno
from sklearn.model_selection import train_test_split as tst
import geopandas as gpd
import branca.colormap as cm
from matplotlib.colors import ListedColormap, LinearSegmentedColormap
import matplotlib.colors
import folium
from folium.plugins import FloatImage
from sklearn.preprocessing import StandardScaler
from sklearn.metrics import r2_score


pd.set_option('display.max_columns', 75)
pd.set_option('display.max_rows', 75)


main_df = pd.read_csv("data/main.csv")
original_df = main_df.copy()
main_df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 11972 entries, 0 to 11971
Data columns (total 74 columns):
 #   Column                                        Non-Null Count  Dtype  
---  ------                                        --------------  -----  
 0   id                                            11972 non-null  int64  
 1   listing_url                                   11972 non-null  object 
 2   scrape_id                                     11972 non-null  int64  
 3   last_scraped                                  11972 non-null  object 
 4   name                                          11972 non-null  object 
 5   description                                   11808 non-null  object 
 6   neighborhood_overview                         7059 non-null   object 
 7   picture_url                                   11971 non-null  object 
 8   host_id                                       11972 non-null  int64  
 9   host_url                                      11972 non-null  object 
 10  host_name                                     11969 non-null  object 
 11  host_since                                    11969 non-null  object 
 12  host_location                                 11954 non-null  object 
 13  host_about                                    7293 non-null   object 
 14  host_response_time                            8523 non-null   object 
 15  host_response_rate                            8523 non-null   object 
 16  host_acceptance_rate                          9110 non-null   object 
 17  host_is_superhost                             11969 non-null  object 
 18  host_thumbnail_url                            11969 non-null  object 
 19  host_picture_url                              11969 non-null  object 
 20  host_neighbourhood                            10254 non-null  object 
 21  host_listings_count                           11969 non-null  float64
 22  host_total_listings_count                     11969 non-null  float64
 23  host_verifications                            11972 non-null  object 
 24  host_has_profile_pic                          11969 non-null  object 
 25  host_identity_verified                        11969 non-null  object 
 26  neighbourhood                                 7059 non-null   object 
 27  neighbourhood_cleansed                        11972 non-null  int64  
 28  neighbourhood_group_cleansed                  0 non-null      float64
 29  latitude                                      11972 non-null  float64
 30  longitude                                     11972 non-null  float64
 31  property_type                                 11972 non-null  object 
 32  room_type                                     11972 non-null  object 
 33  accommodates                                  11972 non-null  int64  
 34  bathrooms                                     0 non-null      float64
 35  bathrooms_text                                11956 non-null  object 
 36  bedrooms                                      11261 non-null  float64
 37  beds                                          11822 non-null  float64
 38  amenities                                     11972 non-null  object 
 39  price                                         11972 non-null  object 
 40  minimum_nights                                11972 non-null  int64  
 41  maximum_nights                                11972 non-null  int64  
 42  minimum_minimum_nights                        11971 non-null  float64
 43  maximum_minimum_nights                        11971 non-null  float64
 44  minimum_maximum_nights                        11971 non-null  float64
 45  maximum_maximum_nights                        11971 non-null  float64
 46  minimum_nights_avg_ntm                        11971 non-null  float64
 47  maximum_nights_avg_ntm                        11971 non-null  float64
 48  calendar_updated                              0 non-null      float64
 49  has_availability                              11972 non-null  object 
 50  availability_30                               11972 non-null  int64  
 51  availability_60                               11972 non-null  int64  
 52  availability_90                               11972 non-null  int64  
 53  availability_365                              11972 non-null  int64  
 54  calendar_last_scraped                         11972 non-null  object 
 55  number_of_reviews                             11972 non-null  int64  
 56  number_of_reviews_ltm                         11972 non-null  int64  
 57  number_of_reviews_l30d                        11972 non-null  int64  
 58  first_review                                  9026 non-null   object 
 59  last_review                                   9026 non-null   object 
 60  review_scores_rating                          9026 non-null   float64
 61  review_scores_accuracy                        8954 non-null   float64
 62  review_scores_cleanliness                     8954 non-null   float64
 63  review_scores_checkin                         8953 non-null   float64
 64  review_scores_communication                   8954 non-null   float64
 65  review_scores_location                        8952 non-null   float64
 66  review_scores_value                           8952 non-null   float64
 67  license                                       0 non-null      float64
 68  instant_bookable                              11972 non-null  object 
 69  calculated_host_listings_count                11972 non-null  int64  
 70  calculated_host_listings_count_entire_homes   11972 non-null  int64  
 71  calculated_host_listings_count_private_rooms  11972 non-null  int64  
 72  calculated_host_listings_count_shared_rooms   11972 non-null  int64  
 73  reviews_per_month                             9026 non-null   float64
dtypes: float64(24), int64(18), object(32)
memory usage: 6.8+ MB


main_df.head(5)


#drop columns that have no potential of giving info
main_df.drop(['picture_url', 'host_url', 'id','scrape_id', 'last_scraped', 'host_id',
              'host_thumbnail_url', 'host_picture_url', 'calendar_last_scraped', 'host_about', 'license','listing_url', 
              'name', 'description','neighborhood_overview', 'host_verifications', 'host_name',
              'neighbourhood_group_cleansed', 'bathrooms','calendar_updated', 'license'], axis=1, inplace=True)


#Get percentage missing from each predictor
missing_count = main_df.isnull().sum()
perc_missing = (missing_count * 100)/len(main_df)
missing_df = pd.DataFrame({'Column': main_df.columns, 'Count': missing_count, '%': perc_missing})
missing_df.sort_values(by='%', ascending=False).reset_index(drop=True)


#Drop unecessary review columns
main_df.drop(['review_scores_value', 'review_scores_location', 'review_scores_checkin', 'review_scores_communication',
          'review_scores_communication', 'review_scores_cleanliness','review_scores_accuracy'], axis=1, inplace=True)


#Split into training and test sets
train_df, test_df = tst(main_df, test_size=0.10, random_state=15413)
df_list = [train_df, test_df]

print('The amount of listings in the training set is %d' % len(train_df) )
print('The amount of listings in the test set is %d' % len(test_df) )

The amount of listings in the training set is 10774
The amount of listings in the test set is 1198


#Palette for all plots
colors = ['orange', 'red', 'gold', 'yellow', 'brown', 'black', 'yellowgreen']

#Helper function to visualize categorical distribution
def get_normalized_count_plot(df, col, labels=None, ax=None):
    """
        get normalized count for a dataframe column
        
        df: dataframe where column is located
        col: predictor to visualize
        ax: the axis for pot (subplots)
    """
    #Get normalized count of each level of predictor
    norm_df = df[col].value_counts(normalize=True).mul(100).reset_index().rename(
        columns={'index': col, col: '%'})
    
    #Plot the distribution of predictors. 
    # Plot as subplot if indicated.
    if ax != None:
        sns.barplot(x=col,y='%', data=norm_df, ax=ax, palette=sns.color_palette(colors))
        ax.set_title(f"Percentage Distribution of '{col}'")
        
        if labels != None:
            label_ticks = np.arange(len(df[col].unique()))
            ax.set_xlabel(label_ticks)
        else:
            ax.set_xlabel(None)
        
    else:
        sns.barplot(x=col,y='%', data=norm_df,  palette=sns.color_palette(colors))
        
        plt.title(f"Percentage Distribution of '{col}'")
        if labels != None:
            label_ticks = np.arange(len(df[col].unique()))
            plt.xticks(ticks = [0,1], labels=labels)
        else:
            plt.xlabel(None)
            
        plt.show()


def get_host_tenure(date_compiled, form, dates):
    """
        create column that counts the number of days a host has been with Airbnb
        
        date_compiled (str): the date the data was collected
        form (str): the format for datetime
        dates (Series): the hosts date they joined Airbnb
    """
    
    #convert both date texts to datetime
    dates = pd.to_datetime(dates)
    date_compiled = pd.to_datetime(date_compiled, format=form)
    
    #Get the number of days
    dates = (date_compiled - dates).dt.days
    return dates


train_df[train_df['host_since'].isnull()]


DATE_COMPILED = "2022-03-20"

#Convert the host_since column to number of days in both dataframes
for df_temp in df_list:
    df_temp.dropna(subset='host_since', inplace=True)
    df_temp['host_since'] = get_host_tenure(DATE_COMPILED,"%Y-%m-%d", df_temp['host_since'])
    df_temp.rename(columns={'host_since': 'host_tenure'},inplace=True)


sns.histplot(x='host_tenure', data=train_df, color='orange')
plt.xlabel('Days')

Text(0.5, 0, 'Days')


# Get distribution of if host is in Texas
df_host_loc = train_df.copy()
df_host_loc.dropna(subset='host_location', inplace=True)
df_host_loc['in_texas'] = np.where(df_host_loc['host_location'].str.contains('Texas', case=False), 1, 0)

get_normalized_count_plot(df_host_loc, 'in_texas', labels=['False', 'True'])


#Create predictor if host in texas
for df1 in df_list:
    df1['in_texas'] = np.where(df1['host_location'].str.contains('Texas', case=False), 'Yes', 'No')
    df1.loc[df1['host_location'].isnull(), 'in_texas'] = 'unknown'
    df1.drop('host_location', axis=1, inplace=True)


get_normalized_count_plot(train_df, 'in_texas')


#Get count of rows where both host_response_time and host_response_rate are missing
time_rate_miss =train_df[
    train_df.loc[:, ['host_response_time', 
                                          'host_response_rate']] \
    .isnull().sum(axis=1) ==2].shape[0]

#Get missing counts where host_response_time is missing
time_miss = train_df[train_df['host_response_time'].isnull()].shape[0]

#Get missing counts where host_response_rate is missing
rate_miss = train_df[train_df['host_response_rate'].isnull()].shape[0]

print("Percentage of 'host_response_time' where 'host_response_rate is also missing: %d'" % 
      (time_rate_miss / time_miss))
print("Percentage of 'host_response_rate' where 'host_response_time is also missing: %d'" % 
      (time_rate_miss / time_miss))

Percentage of 'host_response_time' where 'host_response_rate is also missing: 1'
Percentage of 'host_response_rate' where 'host_response_time is also missing: 1'


get_normalized_count_plot(train_df, 'host_response_time')


#After adding 'unknown' as part of the distribution
for df1 in df_list:
    df1['host_response_time'].fillna('unknown', inplace=True)

plt.figure(figsize=(7, 7))
get_normalized_count_plot(train_df, 'host_response_time')


#Remove percentage sign from string and convert to numeric
for df1 in df_list:
    df1['host_response_rate'] = pd.to_numeric(df1['host_response_rate'].str.strip('%'))

sns.histplot(x='host_response_rate', data=train_df, stat='percent', color='orange')

<AxesSubplot:xlabel='host_response_rate', ylabel='Percent'>


#bin the predictor into 4 different levels
for df1 in df_list:
    df1['host_response_rate'] =  pd.cut(df1['host_response_rate'], 
                                        bins=[0, 90, 99, 100], 
           labels=['Less than 90', '90-99', '100']).astype('str'
                                        ).str.replace('nan', 'unknown')

train_df['host_response_rate'].value_counts(normalize=True)

100             0.563643
unknown         0.297001
90-99           0.073159
Less than 90    0.066196
Name: host_response_rate, dtype: float64


#Remove percentage sign from string and convert to numeric
for df1 in df_list:
    df1['host_acceptance_rate'] = pd.to_numeric(df1['host_acceptance_rate'].str.strip('%'))

sns.histplot(x='host_acceptance_rate', data=train_df, stat='percent', color='orange')

<AxesSubplot:xlabel='host_acceptance_rate', ylabel='Percent'>


#bin the predictor into 6 different levels
for df1 in df_list:
    df1['host_acceptance_rate'] = pd.cut(df1['host_acceptance_rate'], 
                                         bins=[0, 70, 80, 90, 99, 100], 
           labels=['Less than 70', '70-80', '81-90', '91-99', '100']) \
    .astype('str').str.replace('nan', 'unknown')


plt.figure(figsize=(7, 7))
get_normalized_count_plot(train_df, 'host_acceptance_rate')


for df1 in df_list:
    df1.drop('host_neighbourhood', axis=1, inplace=True)


#Get the count of lisitngs where both predictors aren't equal
dif_listing_count = sum(train_df['host_total_listings_count'] != train_df['host_listings_count'])
print("The amount of listings where the host_total_listings_count is not the same as the host_listings_count: ",
     dif_listing_count)

The amount of listings where the host_total_listings_count is not the same as the host_listings_count:  0


#Drop unecessary columns
for df1 in df_list:
    df1.drop(['host_total_listings_count', 'calculated_host_listings_count', 'calculated_host_listings_count_entire_homes',
              'calculated_host_listings_count_private_rooms', 'calculated_host_listings_count_shared_rooms'], inplace=True,
            axis=1)


#Remove neighbourhood, change to categorical type and change to appropriate name
for df1 in df_list:
    df1.drop('neighbourhood', axis=1, inplace=True)
    df1['neighbourhood_cleansed'] = pd.Categorical(df1['neighbourhood_cleansed'])
    df1.rename(columns={'neighbourhood_cleansed': 'zipcode'}, inplace=True)


get_normalized_count_plot(train_df, 'room_type')


 for df1 in df_list:
    # Change Hotel room to a private room type
    df1.loc[df1['room_type'] == 'Hotel room', 'room_type'] = 'Private room'
    
    #Change free form text to categories for property_type
    df1.loc[df1['property_type'].str.contains(
        'home|house|cottage|bungalow|villa|chalet|guest suite| casa ', case=False), 'property_type'] = 'House'
    df1.loc[df1['property_type'].str.contains(
        'rental unit', case=False), 'property_type'] = 'Rental Unit'
    df1.loc[df1['property_type'].str.contains(
        'condominium | condo|apartment', case=False), 'property_type'] = 'Apartment'
    df1.loc[df1['property_type'].str.contains(
        'loft', case=False), 'property_type'] = 'Loft'
    df1.loc[df1['property_type'].str.contains(
        'hotel|resort|bed and breakfast', case=False), 'property_type'] = 'Hotel'
    df1.loc[~df1['property_type'].isin(
        ['House', 'Rental Unit', 'Apartment', 'Loft', 'Hotel']), 'property_type'] = 'Other'


plt.figure(figsize=(7, 7))
get_normalized_count_plot(train_df, 'property_type')


#Extract bathroom count from string. Change to float.
for df1 in df_list:
    df1['bathrooms'] = df1['bathrooms_text'].str.extract(r'(\d*\.?\d+)').astype(float)
    df1.loc[df1['bathrooms_text'].str.contains('half-bath', case=False, na=False), 'bathrooms'] = 0.5
    df1.drop('bathrooms_text', axis=1, inplace=True)

median_bath = train_df['bathrooms'].median()

#Fill missing values with median bathroom count
for df1 in df_list:
    df1['bathrooms'].fillna(median_bath, inplace=True)


get_normalized_count_plot(train_df, 'bathrooms')


def clean_list_as_string(val):
    """
        function for lambda to change from list in a string to list of strings
        
        val: string of the list
    """
    #remove unecessary brackets and double quotes from string
    val = val.strip("[]").replace(r"'", "").replace(", ", ",").replace('"', '').split(",")
    
    #remove unecessary beginning and ending whitespace
    val = list(map(str.strip, val))
    return val



def clean_amenities(df, col):
    """
        Clean each amenities string. Change from string of list to list of strings.
        
        col: the column name containing the amenities
    """
    
    df[col] = df[col].apply(clean_list_as_string)
    return df
    
    

def create_amenity_boolean_df(amenity_series, amenity_list):
    """
        Returns a dataframe stating whether each listing contained the specific amenity in their list
        
        amenity_series: series that contains the amenity list for each airbnb listing
        amenity_list: The list of amenities we want as predictors
    """
    
    amenity_dict = dict()
    for amenity in amenity_list:
        amenity = amenity.lower()
        
        #Create dictionary where each amenity is the key and the is a list of booleans. Each index in this list
        # matches with the row index of our original data frame.
        #The apply function iterates over each listing which checks to see if that amenity is in listings 
        # set of amenties. If it is, True is returned.
        amenity_dict[amenity] = amenity_series.apply(lambda x: any(amenity in string.lower() for string in x))
    return pd.DataFrame(amenity_dict)



def add_amenities(df, col, amenity_list):
    """
        Concatenate the boolean_df for the dataframe to the dataframe passed in the function
        
        df: DataFrame where the amenity series is located
        col: column containing the amenities
        amenity_list: The list of amenities we want as predictors
    """
    amenity_series = df[col]
    boolean_df = create_amenity_boolean_df(amenity_series, amenity_list)
    return pd.concat([df, boolean_df], axis=1)


# list of amenities we are concerned with
amenity_try = ['pool', 'wifi', 'kitchen', 'free parking', 'free street parking','hot tub', 'washer'
               ,'air conditioning', 'workspace', 'pet', 'gym']

#Create boolean predictors for each amenity in the 'amenity' column
train_df = clean_amenities(train_df, 'amenities')
train_df = add_amenities(train_df, 'amenities', amenity_try)
train_df.drop('amenities', axis=1, inplace=True)

test_df = clean_amenities(test_df, 'amenities')
test_df = add_amenities(test_df, 'amenities', amenity_try)
test_df.drop('amenities', axis=1, inplace=True)


fig, ax = plt.subplots(4, 3, figsize=(15,20))
get_normalized_count_plot(train_df, 'pool', ax=ax[0,0])
get_normalized_count_plot(train_df, 'wifi', ax=ax[0,1])
get_normalized_count_plot(train_df, 'kitchen', ax=ax[0,2])
get_normalized_count_plot(train_df, 'free parking', ax=ax[1,0])
get_normalized_count_plot(train_df, 'free street parking', ax=ax[1,1])
get_normalized_count_plot(train_df, 'hot tub', ax=ax[1,2])
get_normalized_count_plot(train_df, 'washer', ax=ax[2,0])
get_normalized_count_plot(train_df, 'air conditioning', ax=ax[2,1])
get_normalized_count_plot(train_df, 'workspace', ax=ax[2,2] )
get_normalized_count_plot(train_df, 'pet', ax=ax[3,0])
get_normalized_count_plot(train_df, 'gym', ax=ax[3,1])


train_df.head()


df_list = [train_df, test_df]
for df1 in df_list:
    df1.drop(['wifi', 'air conditioning', 'pet'], axis=1, inplace=True)


#Change price from a string to a float
train_df['price'] = train_df['price'].str.replace("\$|,", "").astype('float')
train_df = train_df[train_df['price'] >= 10]

test_df['price'] = test_df['price'].str.replace("\$|,", "").astype('float')
test_df = test_df[test_df['price'] >= 10]


train_df[train_df['price'] >= 10000]


plt.figure(figsize=(7, 7))
sns.histplot(x='price', data=train_df, color='orange')

<AxesSubplot:xlabel='price', ylabel='Count'>


train_df.loc[train_df['price'] == 20000, 'price'] = 10000


plt.figure(figsize=(7, 7))
sns.histplot(x='price', data=train_df, color='orange')

<AxesSubplot:xlabel='price', ylabel='Count'>


dif_listing_count = (sum((train_df['minimum_nights'] != train_df['minimum_minimum_nights'])) / train_df.shape[0]) * 100
print('The % of listings where minimum_nights does not equal minimum_minimum_nights is', dif_listing_count)

dif_listing_count = (sum((train_df['maximum_nights'] != train_df['maximum_maximum_nights'])) / train_df.shape[0]) * 100
print('The % of listings where maximum_nights does not equal maximum_aximum_nights is', dif_listing_count)

The % of listings where minimum_nights does not equal minimum_minimum_nights is 7.763744427934621
The % of listings where maximum_nights does not equal maximum_aximum_nights is 20.533060921248143


#Drop uncessary minimum_features
df_list = [train_df, test_df]
for df1 in df_list:
    df1.drop(['minimum_minimum_nights', 'maximum_minimum_nights', 'minimum_maximum_nights', 'maximum_maximum_nights',
              'minimum_nights_avg_ntm', 'maximum_nights_avg_ntm'], axis=1, inplace=True)


#Get distribution of minimum nights
train_df['minimum_nights'].value_counts(normalize=True).head(5)

2     0.334881
1     0.309621
3     0.122493
30    0.106798
4     0.024981
Name: minimum_nights, dtype: float64


#Look at distribution of minimum nights
fig, ax = plt.subplots(1,3, figsize=(20,5))
sns.boxplot(x="minimum_nights", data=train_df, ax=ax[0], color='orange')
ax[0].set_title("minimum_nights")

sns.boxplot(x="minimum_nights", data=train_df[train_df['minimum_nights'] < 100], ax=ax[1], color='orange')
ax[1].set_title("minimum_nights (< 100 Nights)")

sns.boxplot(x="minimum_nights", data=train_df[train_df['minimum_nights'] < 10], ax=ax[2], color='orange')
ax[2].set_title("minimum_nights (<10)")

Text(0.5, 1.0, 'minimum_nights (<10)')


#Distribution of maximum nights
sns.boxplot(x="maximum_nights", data=train_df, color='orange')

<AxesSubplot:xlabel='maximum_nights'>


sns.histplot(x="maximum_nights", data=train_df, color='orange')

<AxesSubplot:xlabel='maximum_nights', ylabel='Count'>


train_df['has_availability'].value_counts(normalize=True).head(5)

t    0.873328
f    0.126672
Name: has_availability, dtype: float64


train_df['availability_30'].value_counts(normalize=True).head(5)

0     0.408432
30    0.054978
10    0.031296
1     0.027953
8     0.026653
Name: availability_30, dtype: float64


train_df['availability_60'].value_counts(normalize=True).head(5)

0     0.346025
60    0.053027
10    0.016995
40    0.016159
31    0.015695
Name: availability_60, dtype: float64


train_df['availability_90'].value_counts(normalize=True).head(5)

0     0.316029
90    0.051542
83    0.012630
40    0.012444
70    0.012444
Name: availability_90, dtype: float64


train_df['availability_365'].value_counts(normalize=True).head(5)

0      0.281575
365    0.044112
364    0.006408
345    0.005944
70     0.005944
Name: availability_365, dtype: float64


for df1 in df_list:
    df1.drop(['availability_30', 'availability_60', 'availability_90', 'availability_365'], axis=1, inplace=True)


print('Correlation between # of reviews and # of reviews the last 12 months: ',
      train_df['number_of_reviews'].corr(train_df['number_of_reviews_ltm']))
print('Correlation between # of reviews and # of reviews the last 30 days: ',
      train_df['number_of_reviews'].corr(train_df['number_of_reviews_l30d']))
print('Correlation between # of reviews and # of reviews per months ',
      train_df['number_of_reviews'].corr(train_df['reviews_per_month']))

Correlation between # of reviews and # of reviews the last 12 months:  0.6883637747838995
Correlation between # of reviews and # of reviews the last 30 days:  0.47224232715841225
Correlation between # of reviews and # of reviews per months  0.5909687011603469


for df1 in df_list:
    df1.drop(['number_of_reviews_ltm', 'reviews_per_month'], axis=1, inplace=True)


train_df["number_of_reviews_l30d"].hist(grid=False, color='orange')

<AxesSubplot:>


train_df["number_of_reviews"].hist(grid=False, color='orange')

<AxesSubplot:>


for df1 in df_list:
    df1.drop(['first_review', 'last_review'], axis=1, inplace=True)


#Get non-nul reviews
df_review_rating = train_df[train_df['review_scores_rating'].notnull()]

#Get % of reviews greater than 4.5
df_review_rating[df_review_rating['review_scores_rating'] > 4.5].shape[0] / df_review_rating.shape[0]

0.8774582104228122


train_df['review_scores_rating'].hist(color='orange')

<AxesSubplot:>


#Appropriate binning for heavily skewed predictor
for df1 in df_list:
    df1['review_scores_rating'] = pd.cut(df1['review_scores_rating'], bins=[0, 4, 4.6, 4.8, 4.99, 5], 
           labels=['Less than 4', '4-4.6', '4.6-4.8', '4.8-4.99', '5']).astype('str')
    df1['review_scores_rating'] = df1['review_scores_rating'].str.replace('nan', 'unknown')


get_normalized_count_plot(train_df, 'review_scores_rating')


#Fill missing values with median of bedrooms
bed_median = train_df['beds'].median()
bedroom_median = train_df['bedrooms'].median()

for df1 in df_list:
    df1['beds'].fillna(bed_median, inplace=True)
    df1['bedrooms'].fillna(bedroom_median, inplace=True)


#Plot distribution of beds
fig, ax = plt.subplots(1,2, figsize=(20,5))
train_df['beds'].hist(ax=ax[0], color='orange')
ax[0].set_title('# of Beds')

train_df[train_df['beds'] < 20]['beds'].hist(ax=ax[1], color='orange')
ax[1].set_title('# of Beds (<20)')

Text(0.5, 1.0, '# of Beds (<20)')


#Plot distribution of bedrooms
fig, ax = plt.subplots(1,2, figsize=(20,5))
train_df['bedrooms'].hist(ax=ax[0], color='orange')
ax[0].set_title('# of Bedrooms')

train_df[train_df['bedrooms'] < 10]['bedrooms'].hist(ax=ax[1], color='orange')
ax[1].set_title('# of Bedrooms (<10)')

Text(0.5, 1.0, '# of Bedrooms (<10)')


train_df.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 10768 entries, 5022 to 8809
Data columns (total 34 columns):
 #   Column                  Non-Null Count  Dtype   
---  ------                  --------------  -----   
 0   host_tenure             10768 non-null  int64   
 1   host_response_time      10768 non-null  object  
 2   host_response_rate      10768 non-null  object  
 3   host_acceptance_rate    10768 non-null  object  
 4   host_is_superhost       10768 non-null  object  
 5   host_listings_count     10768 non-null  float64 
 6   host_has_profile_pic    10768 non-null  object  
 7   host_identity_verified  10768 non-null  object  
 8   zipcode                 10768 non-null  category
 9   latitude                10768 non-null  float64 
 10  longitude               10768 non-null  float64 
 11  property_type           10768 non-null  object  
 12  room_type               10768 non-null  object  
 13  accommodates            10768 non-null  int64   
 14  bedrooms                10768 non-null  float64 
 15  beds                    10768 non-null  float64 
 16  price                   10768 non-null  float64 
 17  minimum_nights          10768 non-null  int64   
 18  maximum_nights          10768 non-null  int64   
 19  has_availability        10768 non-null  object  
 20  number_of_reviews       10768 non-null  int64   
 21  number_of_reviews_l30d  10768 non-null  int64   
 22  review_scores_rating    10768 non-null  object  
 23  instant_bookable        10768 non-null  object  
 24  in_texas                10768 non-null  object  
 25  bathrooms               10768 non-null  float64 
 26  pool                    10768 non-null  bool    
 27  kitchen                 10768 non-null  bool    
 28  free parking            10768 non-null  bool    
 29  free street parking     10768 non-null  bool    
 30  hot tub                 10768 non-null  bool    
 31  washer                  10768 non-null  bool    
 32  workspace               10768 non-null  bool    
 33  gym                     10768 non-null  bool    
dtypes: bool(8), category(1), float64(7), int64(6), object(12)
memory usage: 2.5+ MB


train_df.head(5)


train_df.to_csv('data/train_df')
test_df.to_csv('data/test_df')


print('The # of listings with a price above $1000: ', train_df[train_df['price'] > 800].shape[0])

The # of listings with a price above $1000:  817


price_mean = train_df['price'].mean()

plt.figure(figsize=(20,4))
train_df.price.hist(bins=100, range=(0,800), color='orange')
plt.axvline(price_mean, color='orange')
plt.annotate('<- mean', xy= (price_mean,300), fontsize=20, color='black')
plt.title('Airbnb prices in Austin, Texas')
plt.xlabel('Price($)')
plt.ylabel('# of listings')

Text(0, 0.5, '# of listings')


df2 = train_df.copy()
df2['host_tenure_bins'] = pd.cut(df2['host_tenure'], 
                                 bins=[0,500, 1000, 1500, 2000, 2500, 3000, 3500, 4000, 4500, 5000])
fig, ax = plt.subplots(1,2, figsize=(20,5), sharey=True)

df2[df2['price'] < 500].groupby('host_tenure_bins')['host_listings_count'].median().plot.bar(color='orange', ax=ax[0])
ax[0].set_title('Median # of Listings Count by Host Tenure (Less than $500)')
ax[0].set_ylabel('Host Listings Count')

df2[df2['price'] > 500].groupby('host_tenure_bins')['host_listings_count'].median().plot.bar(color='red', ax=ax[1])
ax[1].set_title('Median # of Listings Count by Host Tenure (More than $500)')
ax[1].set_ylabel('Host Listings Count')

Text(0, 0.5, 'Host Listings Count')


plt.figure(figsize=(5,7))
df2.groupby('host_response_time')['price'].mean().plot.bar(color='orange')
plt.title('Price by Hosts Response Time')

Text(0.5, 1.0, 'Price by Hosts Response Time')


plt.figure(figsize=(20,5))
df2 = train_df[train_df['price'] < 1000]
sns.boxplot(df2['price'], df2['host_is_superhost'], orient="h", palette= {"t": "orange", "f": "red"})

<AxesSubplot:xlabel='price', ylabel='host_is_superhost'>


get_normalized_count_plot(train_df, 'host_has_profile_pic')


for df1 in df_list:
    df1.drop('host_has_profile_pic', axis=1, inplace=True)


plt.figure(figsize=(20,10))
sns.pointplot(x='property_type', y='price', hue='room_type', data=train_df, estimator=np.median,
            palette= {"Entire home/apt": "orange", "Private room": "red", "Shared room": "yellow"})
plt.title('Price of listing by Room and Property Type')

Text(0.5, 1.0, 'Price of listing by Room and Property Type')


fig, ax = plt.subplots(3,1, figsize=(20,10))
sns.barplot(x='accommodates', y='price', data=train_df, ax=ax[0], estimator=np.median, palette = colors)
ax[0].set_title('Accommodates vs Price')

sns.barplot(x='bedrooms', y='price', data=train_df, ax=ax[1], estimator=np.median, palette=colors)
ax[1].set_title('Bedrooms vs Price')

sns.barplot(x='beds', y='price', data=train_df, ax=ax[2], estimator=np.median, palette=colors)
ax[2].set_title('Beds vs Price')
fig.tight_layout()


df= train_df.copy()

AUSTIN_COORD = [30.2672, -97.7431]
quantiles = np.quantile(df['price'], q=[0.0, 0.2, 0.4, 0.6, 0.8, 1])
print(quantiles)
df['price_cut'] = pd.cut(x=df['price'], bins=quantiles, labels=np.arange(1,6))
df['price_cut'] = pd.to_numeric(df['price_cut'])
df['price_cut']. fillna(1, inplace=True)
print(df['price_cut'].unique())

cmap = cm.LinearColormap(colors=["darkorange", "gold", "lawngreen", "lightseagreen", 'blue'],
                        vmin=1, vmax=5)
austin_map = folium.Map(location = AUSTIN_COORD, width=950, height=550,
                       zoom_start=10.5, tiles= 'cartodb positron')

for lat, long, price in zip(df['latitude'], df['longitude'], df['price_cut']):
    folium.Circle(
        location = [lat, long],
        radius=5,
        fill = True,
        color = cmap(price)
    ). add_to(austin_map)

austin_map.add_child(cmap)

[1.0000e+01 1.0000e+02 1.6600e+02 2.5000e+02 4.2660e+02 1.4795e+04]
[1. 5. 2. 3. 4.]


plt.figure(figsize=(40,15))
sns.barplot(x='zipcode', y='price', data=train_df, palette=colors)

<AxesSubplot:xlabel='zipcode', ylabel='price'>


for df1 in df_list:
    df1.drop(['latitude', 'longitude'], inplace=True, axis=1)


train_df.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 10768 entries, 5022 to 8809
Data columns (total 31 columns):
 #   Column                  Non-Null Count  Dtype   
---  ------                  --------------  -----   
 0   host_tenure             10768 non-null  int64   
 1   host_response_time      10768 non-null  object  
 2   host_response_rate      10768 non-null  object  
 3   host_acceptance_rate    10768 non-null  object  
 4   host_is_superhost       10768 non-null  object  
 5   host_listings_count     10768 non-null  float64 
 6   host_identity_verified  10768 non-null  object  
 7   zipcode                 10768 non-null  category
 8   property_type           10768 non-null  object  
 9   room_type               10768 non-null  object  
 10  accommodates            10768 non-null  int64   
 11  bedrooms                10768 non-null  float64 
 12  beds                    10768 non-null  float64 
 13  price                   10768 non-null  float64 
 14  minimum_nights          10768 non-null  int64   
 15  maximum_nights          10768 non-null  int64   
 16  has_availability        10768 non-null  object  
 17  number_of_reviews       10768 non-null  int64   
 18  number_of_reviews_l30d  10768 non-null  int64   
 19  review_scores_rating    10768 non-null  object  
 20  instant_bookable        10768 non-null  object  
 21  in_texas                10768 non-null  object  
 22  bathrooms               10768 non-null  float64 
 23  pool                    10768 non-null  bool    
 24  kitchen                 10768 non-null  bool    
 25  free parking            10768 non-null  bool    
 26  free street parking     10768 non-null  bool    
 27  hot tub                 10768 non-null  bool    
 28  washer                  10768 non-null  bool    
 29  workspace               10768 non-null  bool    
 30  gym                     10768 non-null  bool    
dtypes: bool(8), category(1), float64(5), int64(6), object(11)
memory usage: 2.2+ MB


# train_df.to_csv('data/train_df')
# test_df.to_csv('data/test_df')


for df1 in df_list:
    df1['price'] = df1['price'].astype('int64')
    df1['pool'] = df1['pool'].astype('category')
    df1['kitchen'] = df1['kitchen'].astype('category')
    df1['free parking'] = df1['free parking'].astype('category')
    df1['free street parking'] = df1['free street parking'].astype('category')
    df1['hot tub'] = df1['hot tub'].astype('category')
    df1['washer'] = df1['washer'].astype('category')
    df1['workspace'] = df1['workspace'].astype('category')
    df1['gym'] = df1['gym'].astype('category')


y_train, y_test = train_df['price'], test_df['price']
x_train = train_df.drop('price', axis=1)
x_test = test_df.drop('price', axis=1)


x_train, x_valid, y_train, y_valid = tst(x_train, y_train, test_size=0.20, random_state=15444)


plt.figure(figsize=(10,10))
corr = x_train.corr()

mask = np.zeros_like(corr, dtype=np.bool)
mask[np.triu_indices_from(mask) ] = True

sns.heatmap(corr, mask=mask, center=0, square=True, linewidths=0.5, cbar_kws={"shrink": 0.5})

<AxesSubplot:>


x_train.drop(['bedrooms', 'beds', 'bathrooms'], axis=1, inplace=True)
x_valid.drop(['bedrooms', 'beds', 'bathrooms'], axis=1, inplace=True)
x_test.drop(['bedrooms', 'beds', 'bathrooms'], axis=1, inplace=True)


df_concat = pd.concat([x_train, x_valid, x_test], axis=0)
df_concat['zipcode'] = df_concat['zipcode'].astype('category')
df_concat = pd.get_dummies(df_concat, drop_first=True)
df_concat.columns = df_concat.columns.str.replace(' ','_')


valid_end_indice = len(x_train) + len(x_valid)
x_train = df_concat.iloc[:len(x_train)]
x_valid = df_concat[len(x_train):valid_end_indice]
x_test = df_concat.iloc[valid_end_indice:]


x_test


scaler = StandardScaler()
scaler.fit(x_train)
x_train_std = pd.DataFrame(scaler.transform(x_train), columns= x_train.columns)
x_valid_std = pd.DataFrame(scaler.transform(x_valid), columns= x_train.columns)
x_test_std = pd.DataFrame(scaler.transform(x_test), columns= x_train.columns)


#create lists to track R-squared and MSE
fnl_r_squareds = []
fnl_mse = []
fnl_model_name = []


def get_vifs(df):
    """
        Get a VIF for each predictor
        df: a dataframe containing all of the predictors.
    """
    vif = {}
    #Create dictionary of predictor vifs
    for i in range(len(df.columns)):
        vif[df.columns[i]] = variance_inflation_factor(df.values,i)
    return pd.DataFrame(vif.items(), columns=['Predictor', 'Variance Inflation Factor']).sort_values(
        by='Variance Inflation Factor', ascending=False)


import statsmodels.formula.api as smf
import statsmodels.api as sm
from statsmodels.stats.outliers_influence import variance_inflation_factor
from sklearn import linear_model
from sklearn.metrics import mean_squared_error


#Log transform the predictor
y_train_log = np.log(y_train)
y_valid_log = np.log(y_valid)
y_test_log = np.log(y_test)

model1 = sm.OLS(y_train_log, sm.add_constant(x_train)).fit()
print(model1.summary())

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                  price   R-squared:                       0.598
Model:                            OLS   Adj. R-squared:                  0.594
Method:                 Least Squares   F-statistic:                     144.3
Date:                Wed, 12 Oct 2022   Prob (F-statistic):               0.00
Time:                        20:39:19   Log-Likelihood:                -7633.5
No. Observations:                8614   AIC:                         1.544e+04
Df Residuals:                    8525   BIC:                         1.607e+04
Df Model:                          88                                         
Covariance Type:            nonrobust                                         
=========================================================================================================
                                            coef    std err          t      P>|t|      [0.025      0.975]
---------------------------------------------------------------------------------------------------------
const                                     4.9057      0.150     32.632      0.000       4.611       5.200
host_tenure                            4.176e-05   7.25e-06      5.764      0.000    2.76e-05     5.6e-05
host_listings_count                    2.617e-05   1.32e-05      1.986      0.047    3.35e-07     5.2e-05
accommodates                              0.1394      0.003     53.714      0.000       0.134       0.144
minimum_nights                           -0.0018      0.000     -8.098      0.000      -0.002      -0.001
maximum_nights                        -2.503e-05   1.32e-05     -1.893      0.058   -5.09e-05    8.84e-07
number_of_reviews                        -0.0004      0.000     -3.542      0.000      -0.001      -0.000
number_of_reviews_l30d                    0.0046      0.004      1.045      0.296      -0.004       0.013
host_response_time_unknown                0.0858      0.078      1.104      0.270      -0.067       0.238
host_response_time_within_a_day           0.0541      0.134      0.403      0.687      -0.209       0.317
host_response_time_within_a_few_hours    -0.0673      0.135     -0.498      0.618      -0.332       0.197
host_response_time_within_an_hour         0.0604      0.134      0.450      0.653      -0.203       0.323
host_response_rate_90-99                 -0.0694      0.026     -2.646      0.008      -0.121      -0.018
host_response_rate_Less_than_90          -0.0977      0.030     -3.237      0.001      -0.157      -0.039
host_response_rate_unknown               -0.1290      0.154     -0.839      0.402      -0.431       0.173
host_acceptance_rate_70-80               -0.0428      0.035     -1.212      0.226      -0.112       0.026
host_acceptance_rate_81-90                0.0606      0.026      2.325      0.020       0.010       0.112
host_acceptance_rate_91-99               -0.0075      0.019     -0.397      0.691      -0.044       0.029
host_acceptance_rate_Less_than_70         0.0575      0.029      1.979      0.048       0.001       0.114
host_acceptance_rate_unknown             -0.1311      0.032     -4.054      0.000      -0.195      -0.068
host_is_superhost_t                      -0.0005      0.017     -0.031      0.975      -0.035       0.034
host_identity_verified_t                 -0.0520      0.018     -2.868      0.004      -0.087      -0.016
zipcode_78702                            -0.2165      0.036     -6.024      0.000      -0.287      -0.146
zipcode_78703                            -0.0458      0.044     -1.050      0.294      -0.131       0.040
zipcode_78704                            -0.1918      0.034     -5.571      0.000      -0.259      -0.124
zipcode_78705                            -0.3940      0.045     -8.841      0.000      -0.481      -0.307
zipcode_78712                             1.4823      0.592      2.506      0.012       0.323       2.642
zipcode_78717                            -0.9585      0.097     -9.857      0.000      -1.149      -0.768
zipcode_78719                            -0.8211      0.227     -3.616      0.000      -1.266      -0.376
zipcode_78721                            -0.6003      0.053    -11.302      0.000      -0.704      -0.496
zipcode_78722                            -0.4871      0.060     -8.087      0.000      -0.605      -0.369
zipcode_78723                            -0.5593      0.048    -11.734      0.000      -0.653      -0.466
zipcode_78724                            -0.7581      0.074    -10.193      0.000      -0.904      -0.612
zipcode_78725                            -0.7836      0.105     -7.448      0.000      -0.990      -0.577
zipcode_78726                            -0.8186      0.190     -4.315      0.000      -1.190      -0.447
zipcode_78727                            -0.8490      0.068    -12.433      0.000      -0.983      -0.715
zipcode_78728                            -0.7375      0.072    -10.293      0.000      -0.878      -0.597
zipcode_78729                            -0.6752      0.069     -9.781      0.000      -0.810      -0.540
zipcode_78730                            -0.4334      0.118     -3.671      0.000      -0.665      -0.202
zipcode_78731                            -0.4169      0.064     -6.509      0.000      -0.542      -0.291
zipcode_78732                            -0.4480      0.092     -4.862      0.000      -0.629      -0.267
zipcode_78733                            -0.2990      0.091     -3.286      0.001      -0.477      -0.121
zipcode_78734                            -0.4647      0.053     -8.820      0.000      -0.568      -0.361
zipcode_78735                            -0.5528      0.083     -6.642      0.000      -0.716      -0.390
zipcode_78736                            -0.6065      0.090     -6.753      0.000      -0.783      -0.430
zipcode_78737                            -0.6890      0.064    -10.812      0.000      -0.814      -0.564
zipcode_78738                            -0.4709      0.091     -5.186      0.000      -0.649      -0.293
zipcode_78739                            -0.4175      0.123     -3.400      0.001      -0.658      -0.177
zipcode_78741                            -0.5022      0.040    -12.578      0.000      -0.580      -0.424
zipcode_78742                            -0.8179      0.267     -3.068      0.002      -1.341      -0.295
zipcode_78744                            -0.7642      0.056    -13.639      0.000      -0.874      -0.654
zipcode_78745                            -0.6562      0.043    -15.167      0.000      -0.741      -0.571
zipcode_78746                            -0.1829      0.051     -3.557      0.000      -0.284      -0.082
zipcode_78747                            -0.8272      0.094     -8.814      0.000      -1.011      -0.643
zipcode_78748                            -0.7748      0.062    -12.558      0.000      -0.896      -0.654
zipcode_78749                            -0.6975      0.071     -9.758      0.000      -0.838      -0.557
zipcode_78750                            -0.7732      0.105     -7.363      0.000      -0.979      -0.567
zipcode_78751                            -0.6336      0.045    -14.090      0.000      -0.722      -0.545
zipcode_78752                            -0.5112      0.059     -8.689      0.000      -0.627      -0.396
zipcode_78753                            -0.7528      0.069    -10.849      0.000      -0.889      -0.617
zipcode_78754                            -0.8265      0.066    -12.591      0.000      -0.955      -0.698
zipcode_78756                            -0.5877      0.063     -9.301      0.000      -0.712      -0.464
zipcode_78757                            -0.6803      0.057    -12.004      0.000      -0.791      -0.569
zipcode_78758                            -0.6024      0.044    -13.576      0.000      -0.689      -0.515
zipcode_78759                            -0.7433      0.068    -10.856      0.000      -0.878      -0.609
property_type_Hotel                       0.7443      0.072     10.363      0.000       0.604       0.885
property_type_House                       0.3581      0.024     14.617      0.000       0.310       0.406
property_type_Loft                        0.2886      0.063      4.550      0.000       0.164       0.413
property_type_Other                      -0.0887      0.051     -1.733      0.083      -0.189       0.012
property_type_Rental_Unit                 0.0653      0.024      2.721      0.007       0.018       0.112
room_type_Private_room                   -0.6435      0.021    -30.821      0.000      -0.684      -0.603
room_type_Shared_room                    -1.1364      0.068    -16.795      0.000      -1.269      -1.004
has_availability_t                       -0.0430      0.026     -1.660      0.097      -0.094       0.008
review_scores_rating_4.6-4.8              0.0204      0.030      0.690      0.490      -0.038       0.078
review_scores_rating_4.8-4.99             0.0420      0.029      1.446      0.148      -0.015       0.099
review_scores_rating_5                    0.1347      0.027      4.965      0.000       0.081       0.188
review_scores_rating_Less_than_4          0.0829      0.043      1.926      0.054      -0.001       0.167
review_scores_rating_unknown              0.3154      0.028     11.264      0.000       0.261       0.370
instant_bookable_t                       -0.0307      0.016     -1.907      0.056      -0.062       0.001
in_texas_Yes                              0.0166      0.017      0.955      0.340      -0.017       0.051
in_texas_unknown                          0.0882      0.198      0.445      0.656      -0.300       0.477
pool_True                                 0.0847      0.018      4.593      0.000       0.049       0.121
kitchen_True                             -0.0447      0.031     -1.460      0.144      -0.105       0.015
free_parking_True                        -0.0165      0.018     -0.939      0.348      -0.051       0.018
free_street_parking_True                 -0.0219      0.016     -1.361      0.174      -0.054       0.010
hot_tub_True                              0.1970      0.026      7.521      0.000       0.146       0.248
washer_True                               0.0414      0.022      1.895      0.058      -0.001       0.084
workspace_True                           -0.0281      0.015     -1.932      0.053      -0.057       0.000
gym_True                                  0.0284      0.022      1.279      0.201      -0.015       0.072
==============================================================================
Omnibus:                      900.843   Durbin-Watson:                   2.008
Prob(Omnibus):                  0.000   Jarque-Bera (JB):             6396.853
Skew:                           0.231   Prob(JB):                         0.00
Kurtosis:                       7.196   Cond. No.                     2.29e+05
==============================================================================

Notes:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 2.29e+05. This might indicate that there are
strong multicollinearity or other numerical problems.


fitted_vals = model1.predict()
resids = model1.resid

fig, ax = plt.subplots(1,2, figsize=(20,10))

sns.regplot(x=fitted_vals, y=y_train_log, lowess=True, ax=ax[0], line_kws={'color': 'red'})
ax[0].set_title('Observed vs. Predicted Values', fontsize=16)
ax[0].set(xlabel='Predicted', ylabel='Observed')

sns.regplot(x=fitted_vals, y=resids, lowess=True, ax=ax[1], line_kws={'color': 'red'})
ax[1].set_title('Residuals vs. Fitted Values', fontsize=16)
ax[1].set(xlabel='Fitted', ylabel='Residuals')

[Text(0.5, 0, 'Fitted'), Text(0, 0.5, 'Residuals')]


#Create Correlation heatmap
plt.figure(figsize=(15,15))
corr = x_train_std.corr()

mask = np.zeros_like(corr, dtype=np.bool)
mask[np.triu_indices_from(mask) ] = True

sns.heatmap(corr, cmap='YlOrRd', mask=mask, center=0, square=True, linewidths=0.5, cbar_kws={"shrink": 0.5})

<AxesSubplot:>


zipcodes = x_train.loc[:, x_train.columns.str.contains('zipcode')].columns
matrix_cols = x_train_std.drop(zipcodes, axis=1)

plt.figure(figsize=(15,15))
corr = matrix_cols.corr()

mask = np.zeros_like(corr, dtype=np.bool)
mask[np.triu_indices_from(mask) ] = True

sns.heatmap(corr, cmap='YlOrRd', mask=mask, center=0, square=True, linewidths=0.5, cbar_kws={"shrink": 0.5})

<AxesSubplot:>


get_vifs(x_train).head(15)


x_train['pool_True'].corr(np.log(y_train))

0.07078968988925738


x_train['gym_True'].corr(np.log(y_train))

0.01952890606196009


#Drop collinear predictors
x_train.drop(
    ['host_response_time_within_a_few_hours', 'host_response_time_within_an_hour', 'host_response_time_unknown',
    'host_response_time_within_a_day', 'has_availability_t', 'kitchen_True', 'property_type_House', 
     'host_response_rate_unknown', 'gym_True', 'host_is_superhost_t'], axis=1, inplace=True)

x_valid.drop(
    ['host_response_time_within_a_few_hours', 'host_response_time_within_an_hour', 'host_response_time_unknown',
    'host_response_time_within_a_day', 'has_availability_t', 'kitchen_True', 'property_type_House', 
     'host_response_rate_unknown', 'gym_True', 'host_is_superhost_t'], axis=1, inplace=True)

x_test.drop(
    ['host_response_time_within_a_few_hours', 'host_response_time_within_an_hour', 'host_response_time_unknown',
    'host_response_time_within_a_day', 'has_availability_t', 'kitchen_True', 'property_type_House', 
     'host_response_rate_unknown', 'gym_True', 'host_is_superhost_t'], axis=1, inplace=True)


model1 = sm.OLS(y_train_log, sm.add_constant(x_train)).fit()
print(model1.summary())

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                  price   R-squared:                       0.586
Model:                            OLS   Adj. R-squared:                  0.583
Method:                 Least Squares   F-statistic:                     155.1
Date:                Fri, 17 Jun 2022   Prob (F-statistic):               0.00
Time:                        18:34:50   Log-Likelihood:                -7760.5
No. Observations:                8614   AIC:                         1.568e+04
Df Residuals:                    8535   BIC:                         1.624e+04
Df Model:                          78                                         
Covariance Type:            nonrobust                                         
=====================================================================================================
                                        coef    std err          t      P>|t|      [0.025      0.975]
-----------------------------------------------------------------------------------------------------
const                                 5.0001      0.052     96.088      0.000       4.898       5.102
host_tenure                        5.072e-05   7.29e-06      6.954      0.000    3.64e-05     6.5e-05
host_listings_count                 2.97e-05   1.32e-05      2.248      0.025     3.8e-06    5.56e-05
accommodates                          0.1480      0.003     57.934      0.000       0.143       0.153
minimum_nights                       -0.0020      0.000     -8.886      0.000      -0.002      -0.002
maximum_nights                    -2.742e-05   1.33e-05     -2.056      0.040   -5.36e-05   -1.28e-06
number_of_reviews                    -0.0004      0.000     -3.511      0.000      -0.001      -0.000
number_of_reviews_l30d                0.0085      0.004      1.914      0.056      -0.000       0.017
host_response_rate_90-99             -0.0811      0.026     -3.071      0.002      -0.133      -0.029
host_response_rate_Less_than_90      -0.0940      0.028     -3.303      0.001      -0.150      -0.038
host_acceptance_rate_70-80           -0.0588      0.035     -1.665      0.096      -0.128       0.010
host_acceptance_rate_81-90            0.0227      0.026      0.877      0.381      -0.028       0.074
host_acceptance_rate_91-99            0.0019      0.019      0.099      0.921      -0.035       0.039
host_acceptance_rate_Less_than_70     0.0180      0.028      0.651      0.515      -0.036       0.072
host_acceptance_rate_unknown         -0.1857      0.022     -8.612      0.000      -0.228      -0.143
host_identity_verified_t             -0.0627      0.018     -3.445      0.001      -0.098      -0.027
zipcode_78702                        -0.1094      0.035     -3.105      0.002      -0.179      -0.040
zipcode_78703                         0.0232      0.043      0.538      0.591      -0.061       0.108
zipcode_78704                        -0.1036      0.034     -3.071      0.002      -0.170      -0.037
zipcode_78705                        -0.3737      0.044     -8.462      0.000      -0.460      -0.287
zipcode_78712                         1.6068      0.600      2.679      0.007       0.431       2.783
zipcode_78717                        -0.7915      0.098     -8.093      0.000      -0.983      -0.600
zipcode_78719                        -0.6830      0.230     -2.972      0.003      -1.134      -0.233
zipcode_78721                        -0.4848      0.053     -9.193      0.000      -0.588      -0.381
zipcode_78722                        -0.3601      0.060     -5.983      0.000      -0.478      -0.242
zipcode_78723                        -0.4362      0.047     -9.278      0.000      -0.528      -0.344
zipcode_78724                        -0.6104      0.074     -8.213      0.000      -0.756      -0.465
zipcode_78725                        -0.6307      0.106     -5.958      0.000      -0.838      -0.423
zipcode_78726                        -0.6958      0.192     -3.623      0.000      -1.072      -0.319
zipcode_78727                        -0.7392      0.069    -10.771      0.000      -0.874      -0.605
zipcode_78728                        -0.6260      0.072     -8.688      0.000      -0.767      -0.485
zipcode_78729                        -0.5664      0.069     -8.184      0.000      -0.702      -0.431
zipcode_78730                        -0.3426      0.119     -2.868      0.004      -0.577      -0.108
zipcode_78731                        -0.3267      0.064     -5.096      0.000      -0.452      -0.201
zipcode_78732                        -0.3027      0.093     -3.263      0.001      -0.484      -0.121
zipcode_78733                        -0.1158      0.091     -1.275      0.202      -0.294       0.062
zipcode_78734                        -0.3329      0.052     -6.383      0.000      -0.435      -0.231
zipcode_78735                        -0.4936      0.084     -5.866      0.000      -0.659      -0.329
zipcode_78736                        -0.4777      0.090     -5.289      0.000      -0.655      -0.301
zipcode_78737                        -0.5385      0.063     -8.508      0.000      -0.663      -0.414
zipcode_78738                        -0.3649      0.091     -3.993      0.000      -0.544      -0.186
zipcode_78739                        -0.2700      0.124     -2.184      0.029      -0.512      -0.028
zipcode_78741                        -0.4153      0.040    -10.496      0.000      -0.493      -0.338
zipcode_78742                        -0.6879      0.270     -2.546      0.011      -1.217      -0.158
zipcode_78744                        -0.6274      0.056    -11.274      0.000      -0.736      -0.518
zipcode_78745                        -0.5516      0.043    -12.894      0.000      -0.636      -0.468
zipcode_78746                        -0.0662      0.051     -1.292      0.196      -0.167       0.034
zipcode_78747                        -0.6993      0.094     -7.416      0.000      -0.884      -0.514
zipcode_78748                        -0.6270      0.061    -10.200      0.000      -0.748      -0.507
zipcode_78749                        -0.5604      0.071     -7.838      0.000      -0.701      -0.420
zipcode_78750                        -0.6625      0.106     -6.263      0.000      -0.870      -0.455
zipcode_78751                        -0.5539      0.045    -12.438      0.000      -0.641      -0.467
zipcode_78752                        -0.4102      0.059     -6.964      0.000      -0.526      -0.295
zipcode_78753                        -0.6313      0.070     -9.083      0.000      -0.768      -0.495
zipcode_78754                        -0.6912      0.066    -10.519      0.000      -0.820      -0.562
zipcode_78756                        -0.5026      0.063     -7.928      0.000      -0.627      -0.378
zipcode_78757                        -0.5892      0.057    -10.402      0.000      -0.700      -0.478
zipcode_78758                        -0.5121      0.044    -11.527      0.000      -0.599      -0.425
zipcode_78759                        -0.6441      0.069     -9.356      0.000      -0.779      -0.509
property_type_Hotel                   0.5690      0.071      8.064      0.000       0.431       0.707
property_type_Loft                    0.0665      0.062      1.069      0.285      -0.055       0.188
property_type_Other                  -0.3963      0.047     -8.401      0.000      -0.489      -0.304
property_type_Rental_Unit            -0.1470      0.019     -7.808      0.000      -0.184      -0.110
room_type_Private_room               -0.5977      0.021    -28.803      0.000      -0.638      -0.557
room_type_Shared_room                -1.1517      0.068    -16.905      0.000      -1.285      -1.018
review_scores_rating_4.6-4.8          0.0266      0.030      0.888      0.375      -0.032       0.085
review_scores_rating_4.8-4.99         0.0537      0.028      1.899      0.058      -0.002       0.109
review_scores_rating_5                0.1502      0.027      5.527      0.000       0.097       0.204
review_scores_rating_Less_than_4      0.0793      0.044      1.821      0.069      -0.006       0.165
review_scores_rating_unknown          0.3292      0.028     11.712      0.000       0.274       0.384
instant_bookable_t                   -0.0368      0.016     -2.361      0.018      -0.067      -0.006
in_texas_Yes                          0.0192      0.018      1.092      0.275      -0.015       0.054
in_texas_unknown                      0.0575      0.201      0.286      0.775      -0.336       0.451
pool_True                             0.0218      0.017      1.307      0.191      -0.011       0.054
free_parking_True                    -0.0032      0.018     -0.182      0.855      -0.038       0.031
free_street_parking_True              0.0073      0.016      0.455      0.649      -0.024       0.039
hot_tub_True                          0.2251      0.026      8.505      0.000       0.173       0.277
washer_True                           0.0058      0.020      0.288      0.774      -0.034       0.046
workspace_True                       -0.0214      0.015     -1.454      0.146      -0.050       0.007
==============================================================================
Omnibus:                      833.632   Durbin-Watson:                   2.008
Prob(Omnibus):                  0.000   Jarque-Bera (JB):             5412.644
Skew:                           0.214   Prob(JB):                         0.00
Kurtosis:                       6.860   Cond. No.                     2.29e+05
==============================================================================

Notes:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 2.29e+05. This might indicate that there are
strong multicollinearity or other numerical problems.


from mlxtend.feature_selection import SequentialFeatureSelector as SFS
from mlxtend.plotting import plot_sequential_feature_selection as plot_sfs
from sklearn.linear_model import LinearRegression


# Initiate sequential forward selection

lr = LinearRegression()
sfs = SFS(
    lr,
    k_features=x_train.shape[1],
    forward=True,
    cv=3
)
sfs.fit(x_train, y_train_log)


plot_sfs(sfs.get_metric_dict(), kind='std_err', figsize=(20,10))
plt.title('Sequential Forward Selection')
_ = plt.xticks(np.arange(1, x_train.shape[1], ))


print('The feature that contributes the most to the price of an Airbnb is: ', list(sfs.subsets_[1]['feature_names'])[0])
print('The two features that contribute the most to the price of an Airbnb are: ', list(sfs.subsets_[2]['feature_names']))
print('The three features that contribute the most to the price of an Airbnb are: ', list(sfs.subsets_[3]['feature_names']))

The feature that contributes the most to the price of an Airbnb is:  accommodates
The two features that contribute the most to the price of an Airbnb are:  ['accommodates', 'room_type_Private_room']
The three features that contribute the most to the price of an Airbnb are:  ['accommodates', 'room_type_Private_room', 'room_type_Shared_room']


#Get the best 18 feature combination
linear_features = list(sfs.subsets_[18]['feature_names'])
temp = x_train[linear_features]

model1 = sm.OLS(y_train_log, sm.add_constant(temp)).fit()
print(model1.summary())

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                  price   R-squared:                       0.560
Model:                            OLS   Adj. R-squared:                  0.559
Method:                 Least Squares   F-statistic:                     607.6
Date:                Fri, 17 Jun 2022   Prob (F-statistic):               0.00
Time:                        18:35:23   Log-Likelihood:                -8026.7
No. Observations:                8614   AIC:                         1.609e+04
Df Residuals:                    8595   BIC:                         1.623e+04
Df Model:                          18                                         
Covariance Type:            nonrobust                                         
================================================================================================
                                   coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------------------------
const                            4.5704      0.028    166.174      0.000       4.516       4.624
host_tenure                   5.202e-05    6.8e-06      7.653      0.000    3.87e-05    6.53e-05
host_listings_count           6.612e-05   1.26e-05      5.233      0.000    4.14e-05    9.09e-05
accommodates                     0.1467      0.002     58.826      0.000       0.142       0.152
minimum_nights                  -0.0023      0.000     -9.855      0.000      -0.003      -0.002
host_acceptance_rate_unknown    -0.1469      0.016     -8.975      0.000      -0.179      -0.115
zipcode_78702                    0.3286      0.021     15.772      0.000       0.288       0.369
zipcode_78703                    0.4627      0.033     14.200      0.000       0.399       0.527
zipcode_78704                    0.3423      0.019     17.901      0.000       0.305       0.380
zipcode_78746                    0.3785      0.044      8.517      0.000       0.291       0.466
property_type_Hotel              0.8375      0.070     12.009      0.000       0.701       0.974
property_type_Other             -0.4156      0.047     -8.834      0.000      -0.508      -0.323
property_type_Rental_Unit       -0.1476      0.017     -8.738      0.000      -0.181      -0.114
room_type_Private_room          -0.6654      0.020    -32.595      0.000      -0.705      -0.625
room_type_Shared_room           -1.1997      0.068    -17.744      0.000      -1.332      -1.067
review_scores_rating_5           0.1363      0.016      8.484      0.000       0.105       0.168
review_scores_rating_unknown     0.3164      0.018     17.851      0.000       0.282       0.351
free_parking_True               -0.0957      0.017     -5.660      0.000      -0.129      -0.063
hot_tub_True                     0.2913      0.026     11.301      0.000       0.241       0.342
==============================================================================
Omnibus:                      796.617   Durbin-Watson:                   2.004
Prob(Omnibus):                  0.000   Jarque-Bera (JB):             4468.789
Skew:                           0.258   Prob(JB):                         0.00
Kurtosis:                       6.491   Cond. No.                     2.53e+04
==============================================================================

Notes:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 2.53e+04. This might indicate that there are
strong multicollinearity or other numerical problems.


#Get appropriate predictors from validation set
temp_valid = sm.add_constant(x_valid[linear_features])
lin_reg_pred = model1.predict(temp_valid)

#Get MSE and R^2 for validation set
r_squared_lnr = r2_score(y_valid_log, lin_reg_pred)
mse_lnr = mean_squared_error(y_valid_log, lin_reg_pred)

#Append results for final model comparison
fnl_r_squareds.append(r_squared_lnr)
fnl_mse.append(mse_lnr)
fnl_model_name.append('Linear Regression')
print('The MSE of the validation data is: ', mse_lnr)
print('The R^2 of the validation data is: ', r_squared_lnr)

The MSE of the validation data is:  0.3932776361013266
The R^2 of the validation data is:  0.527486096276466


def get_ridge_coefs(cols, ridge_model_ins, sortby ='coef', ascend=False):
    """
        return sorted ridge coefficients and their associated features
        
        cols: names of the features
        ridge_model_ins : instance of a ridge regression model
        ascend If True, will return dataframe sorted with the smallest value first
        sortby: How to sort the dataframe
                - 'coef': sorts dataframe by features coefficient value
                - 'magnitude': sorts by the absolute value of the coefficient
    """
    #get indices of sorted coefficients, smallest to largest
    coefs_temp = ridge_model_ins.coef_
    coefs_inds_sorted = np.argsort(coefs_temp)
    
    #sort features and coefficents from smallest to largest
    coefs_sorted = coefs_temp[coefs_inds_sorted]
    features_sorted = cols[coefs_inds_sorted]
    
    #Place coefficients in Dataframe
    coefficients = pd.DataFrame(zip(features_sorted, coefs_sorted), columns=['feature', 'coef'])
    coefficients['magnitude'] = abs(coefficients['coef'])
    
    return coefficients.sort_values(by=sortby, ascending=ascend)


#Values of our hyperparameter
alpha_list = []

#Store R^2 for different alphas
r_squared = []

#Store R^2 for different alphas
mse = []

#Store coefficients of the model
coefs = []

#Largest and smallest coefficient 
max_coef = None
min_coef = None

for i in range (-2,6):
    #Get alpha
    alpha = 10**i
    
    #Fit Ridge Model
    rm = linear_model.Ridge(alpha=alpha)
    ridge_model = rm.fit(x_train_std, y_train_log)
    
    #make predictions
    preds_ridge = ridge_model.predict(x_train_std)
    
    #get max, and min
    if i == 5:
        feature_df = get_ridge_coefs(x_train_std.columns, ridge_model )
        
        #get min and max features
        min_coef1 = feature_df['feature'].iloc[-1]
        max_coef1 = feature_df['feature'].iloc[0]

    #store values for each alpha
    alpha_list.append(alpha)
    r_squared.append(ridge_model.score(x_train_std, y_train_log))
    mse.append(mean_squared_error(y_train_log, preds_ridge))
    coefs.append(ridge_model.coef_)

#Dataframe of results for alpha
results = pd.DataFrame(zip(alpha_list, r_squared, mse), columns=['Alpha', 'R^2', 'MSE'])
results['Alpha'] = results['Alpha'].astype('str')


#Instantiate flexible figure
fig = plt.figure(figsize=(20,15))
grid = fig.add_gridspec(3, 2)
ax0 = fig.add_subplot(grid[0, 0])
ax1 = fig.add_subplot(grid[0, 1])
ax2 = fig.add_subplot(grid[1:, :])

#Plot R^2
ax0.plot(results["Alpha"], results['R^2'])
ax0.set_title("Ridge  Regression: R^2 by Alpha")

#Plot MSE
ax1.plot(results["Alpha"], results['MSE'])
ax1.set_title("Ridge  Regression: MSE by Alpha")

#Plot Coefficients by alpha
ax2 = plt.gca()
ax2.plot(results["Alpha"], coefs)
ax2.set_xlabel("alpha")
ax2.set_ylabel("weights")

#Add largest and smallest coefficnet annotations
ax2.annotate(text=max_coef1, xy=(0,0.4))
ax2.annotate(text=min_coef1, xy=(0, -0.23))
ax2.set_title("Ridge coefficients as a function of the regularization")

Text(0.5, 1.0, 'Ridge coefficients as a function of the regularization')


#Fit with best Alpha
rm = linear_model.Ridge(alpha=1000)
ridge_model = rm.fit(x_train_std, y_train_log)
preds_ridge = ridge_model.predict(x_valid_std)

#Get R^2 and MSE
final_ridge_r_squared = ridge_model.score(x_valid_std, y_valid_log)
final_mse = mean_squared_error(y_valid_log, preds_ridge)

#Store for final model comparison
fnl_r_squareds.append(final_ridge_r_squared)
fnl_mse.append(final_mse)
fnl_model_name.append('Ridge Regression')

print('The MSE of our final ridge model is ', final_mse)
print('The R^2 of our final model is ', final_ridge_r_squared)

#Table visualization of Coefficients
ridge_coefs = get_ridge_coefs(x_train_std.columns, ridge_model, sortby='magnitude')
ridge_coefs.head(6)

The MSE of our final ridge model is  0.36385915042551825
The R^2 of our final model is  0.562831720416481


import xgboost as xgb


#Fit model
xgb_regressor = xgb.XGBRegressor()
xgb_regressor.fit(x_train, y_train_log)

#Make predictions
xgb_preds = xgb_regressor.predict(x_train)

print("MSE: ", mean_squared_error(y_train_log, xgb_preds))
print("R^2: ", r2_score(y_train_log, xgb_preds))

MSE:  0.13028960249860458
R^2:  0.8481216787556805


lambda_ = []
r_squared = []
mse = []

#Plot of MSE and R^2 of different values of Lambda
for i in range (0,1000, 100):
    xgb_regressor = xgb.XGBRegressor(objective='reg:squarederror', reg_lambda = i)
    xgb_regressor.fit(x_train, y_train_log)
    xgb_preds = xgb_regressor.predict(x_valid)
    
    lambda_.append(i)
    mse.append(mean_squared_error(y_valid_log, xgb_preds))
    r_squared.append(r2_score(y_valid_log, xgb_preds))

fig,ax = plt.subplots(1,2, figsize=(20,7))
ax[0].plot(lambda_, r_squared)
ax[0].set_title("R-squared via Lambda")
ax[0].set_xlabel("Lambda")
ax[0].set_ylabel('R^2')
ax[1].plot(lambda_, mse)
ax[1].set_xlabel("Lambda")
ax[1].set_ylabel('MSE')
ax[1].set_title("MSE via Lambda")
plt.plot()

[]


eta=np.arange(0.1, 0.9, 0.1)
print(eta)
r_squared = []
mse = []

#Plot of MSE and R^2 of different values of the Learning Rate
for i in eta:
    xgb_regressor = xgb.XGBRegressor(objective='reg:squarederror', eta=i)
    xgb_regressor.fit(x_train, y_train_log)
    xgb_preds = xgb_regressor.predict(x_valid)
    
    mse.append(mean_squared_error(y_valid_log, xgb_preds))
    r_squared.append(r2_score(y_valid_log, xgb_preds))

fig,ax = plt.subplots(1,2, figsize=(20,7))
ax[0].plot(eta, r_squared)
ax[0].set_title("R-squared via Learning Rate")
ax[0].set_xlabel("Learning Rate")
ax[0].set_ylabel('R^2')
ax[1].plot(eta, mse)
ax[1].set_xlabel("Learning Rate")
ax[1].set_ylabel('MSE')
ax[1].set_title("MSE via Learning Rate")
plt.plot()

[0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8]

[]


gamma=np.arange(0,10, 1)
r_squared = []
mse = []

#Plot of MSE and R^2 of different values of Gamma
for i in gamma:
    xgb_regressor = xgb.XGBRegressor(objective='reg:squarederror', gamma=i)
    xgb_regressor.fit(x_train, y_train_log)
    xgb_preds = xgb_regressor.predict(x_valid)
    
    mse.append(mean_squared_error(y_valid_log, xgb_preds))
    r_squared.append(r2_score(y_valid_log, xgb_preds))

fig,ax = plt.subplots(1,2, figsize=(20,7))
ax[0].plot(gamma, r_squared)
ax[0].set_title("R-squared via Gamma")
ax[0].set_xlabel("Gamma")
ax[0].set_ylabel('R^2')
ax[1].plot(gamma, mse)
ax[1].set_xlabel("Gamma")
ax[1].set_ylabel('MSE')
ax[1].set_title("MSE via Gamma")
plt.plot()

[]


depth=np.arange(1,50, 5)
r_squared = []
mse = []

#Plot of MSE and R^2 of different values of max_depth
for i in depth:
    xgb_regressor = xgb.XGBRegressor(objective='reg:squarederror', max_depth=i)
    xgb_regressor.fit(x_train, y_train_log)
    xgb_preds = xgb_regressor.predict(x_valid)
    
    mse.append(mean_squared_error(y_valid_log, xgb_preds))
    r_squared.append(r2_score(y_valid_log, xgb_preds))

fig,ax = plt.subplots(1,2, figsize=(20,7))
ax[0].plot(depth, r_squared)
ax[0].set_title("R-squared via max_depth")
ax[0].set_xlabel("Max Depth")
ax[0].set_ylabel('R^2')
ax[1].plot(depth, mse)
ax[1].set_xlabel("Max Depth")
ax[1].set_ylabel('MSE')
ax[1].set_title("MSE via max_depth")
plt.plot()

[]


estimator_counts=np.arange(10,500, 50)
r_squared = []
mse = []

#Plot of MSE and R^2 of different values of n_estimators
for i in estimator_counts:
    xgb_regressor = xgb.XGBRegressor(objective='reg:squarederror', n_estimators=i)
    xgb_regressor.fit(x_train, y_train_log)
    xgb_preds = xgb_regressor.predict(x_valid)
    
    mse.append(mean_squared_error(y_valid_log, xgb_preds))
    r_squared.append(r2_score(y_valid_log, xgb_preds))

fig,ax = plt.subplots(1,2, figsize=(20,7))
ax[0].plot(estimator_counts, r_squared)
ax[0].set_title("R-squared via n_estimatrs")
ax[0].set_xlabel("n_estimators")
ax[0].set_ylabel('R^2')
ax[1].plot(estimator_counts, mse)
ax[1].set_xlabel("n_estimators")
ax[1].set_ylabel('MSE')
ax[1].set_title("MSE via n_estimators")
plt.plot()

[]


#Fit final model
xgb_regressor = xgb.XGBRegressor(objective='reg:squarederror', reg_lambda=125, eta=0.2, gamma=1,
                                max_depth=8, n_estimators=50)
xgb_regressor.fit(x_train, y_train_log)

#Predictions of validation and training set
xgb_preds = xgb_regressor.predict(x_train)
xgb_preds_valid = xgb_regressor.predict(x_valid)

#MSE and R^2 of training set
fnl_mse_xgb = mean_squared_error(y_train_log, xgb_preds)
fnl_r_squared_xgb = r2_score(y_train_log, xgb_preds)

#MSE and R^2 of validation set
fnl_mse_xgb_valid = mean_squared_error(y_valid_log, xgb_preds_valid)
fnl_r_squared_xgb_valid = r2_score(y_valid_log, xgb_preds_valid)


fnl_r_squareds.append(fnl_r_squared_xgb_valid)
fnl_mse.append(fnl_mse_xgb_valid)
fnl_model_name.append('XGB Regressor')

print('The MSE of our final on the Training Data XGB model is ', fnl_mse_xgb)
print('The R^2 of our final on the Training Data XGB model is ', fnl_r_squared_xgb)
print("--------------------------------------------")
print('The MSE of our final on the Validation Data XGB model is ', fnl_mse_xgb_valid)
print('The R^2 of our final on the Validation Data XGB model is ', fnl_r_squared_xgb_valid)

The MSE of our final on the Training Data XGB model is  0.2543476430970742
The R^2 of our final on the Training Data XGB model is  0.7035074763817266
--------------------------------------------
The MSE of our final on the Validation Data XGB model is  0.30197739157688774
The R^2 of our final on the Validation Data XGB model is  0.6371812098324292


from sklearn.model_selection import GridSearchCV


model_metrics_df = pd.DataFrame(zip(fnl_model_name, fnl_r_squareds, fnl_mse), columns = ['Model', 'R^2', 'MSE'])
model_metrics_df


#parameters for GridSearchCV
parameters = {
    
    'reg_lambda': [5, 50, 100, 200],
    'learning_rate': [0.1,0.2,0.3],
    'gamma': [1, 2, 3],
    'max_depth': [5, 10, 15],
    'n_estimators' : [20, 50, 100]
}


#Commented out for timing issues

# clf = GridSearchCV(xgb, parameters, n_jobs=5, cv=3, scoring = 'r2', verbose=2)
# clf.fit(x_train, y_train_log)


#Commented out for timing issues: the params for thes test were shown above

# print(f'The best possible score given the different hyperparameters is: {clf.best_score_}')
# print(f'The best parameters are: {clf.best_params_}')


final_model = xgb.XGBRegressor(objective='reg:squarederror', reg_lambda=5, eta=0.1, gamma=1,
                               max_depth=10, n_estimators=100)
final_model.fit(x_train, y_train_log)

final_model_preds = final_model.predict(x_valid)

fnl_mse = mean_squared_error(y_valid_log, final_model_preds)
fnl_r_squared = r2_score(y_valid_log, final_model_preds)

print('The MSE of our final on the Validation Data XGB model is ', fnl_mse)
print('The R^2 of our final on the Validation Data XGB model is ', fnl_r_squared)

The MSE of our final on the Validation Data XGB model is  0.2936062178998558
The R^2 of our final on the Validation Data XGB model is  0.6472389796870643


#Get feature importance and sort from smallest to Largest
feature_weights = pd.DataFrame(zip(x_train.columns, final_model.feature_importances_), columns=['Feature', 'Importance'])
feature_weights = feature_weights.sort_values('Importance', ascending=False)

plt.figure(figsize=(10,25))
plt.barh(feature_weights['Feature'], feature_weights['Importance'], align='center') 
plt.title("XGBRegression Feature Importance", fontsize=14)
plt.xlabel("Feature importance")
plt.margins(y=0.01)
plt.show()


# list of number of features to loop over
feature_counts = [10, 18, 30, 40, 50, 60, 70, 78]

mse = []
r_squared = []
for i in feature_counts:
    
    #get the subset of features
    cols = feature_weights['Feature'][0:i]
    subset_train = x_train[cols]
    subset_valid = x_valid[cols]
    
    #Fit and get predictions with feature subset
    xgb_regressor =  xgb.XGBRegressor(objective='reg:squarederror', reg_lambda=5, eta=0.1, gamma=1,
                                      max_depth=10, n_estimators=100)
    xgb_regressor.fit(subset_train, y_train_log)
    xgb_preds = xgb_regressor.predict(subset_valid)
    
    #Get MSE and R^2 for each feature subset
    mse.append(mean_squared_error(y_valid_log, xgb_preds))
    r_squared.append(r2_score(y_valid_log, xgb_preds))

#Plot subplots
fig,ax = plt.subplots(1,2, figsize=(20,7))
ax[0].plot(feature_counts, r_squared)
ax[0].set_title("R-squared ")
ax[0].set_xlabel("# of Features")
ax[0].set_ylabel('R^2')
ax[1].plot(feature_counts, mse)
ax[1].set_xlabel("# of Features")
ax[1].set_ylabel('MSE')
ax[1].set_title("MSE ")
plt.plot()

[]


#Create final predictors with chosen subset amount
cols = feature_weights['Feature'][0:30]
subset_train = x_train[cols]
subset_valid = x_valid[cols]

#Fit Final model and get predictions
xgb_final =  xgb.XGBRegressor(objective='reg:squarederror', reg_lambda=5, eta=0.1, gamma=1,
                                  max_depth=10, n_estimators=100)
xgb_final.fit(subset_train, y_train_log)
xgb_preds = xgb_final.predict(subset_valid)


print('The MSE of our final on the Validation Data XGB model is ', mean_squared_error(y_valid_log, xgb_preds))
print('The R^2 of our final on the Validation Data XGB model is ', r2_score(y_valid_log, xgb_preds))

The MSE of our final on the Validation Data XGB model is  0.2944711392678445
The R^2 of our final on the Validation Data XGB model is  0.6461997968439877


x_test = x_test[cols]
test_preds = xgb_final.predict(x_test)

print('The MSE of our final on the Validation Data XGB model is ', mean_squared_error(y_test_log, test_preds))
print('The R^2 of our final on the Validation Data XGB model is ', r2_score(y_test_log, test_preds))

The MSE of our final on the Validation Data XGB model is  0.28863948263390005
The R^2 of our final on the Validation Data XGB model is  0.6557908804179018


from xgboost import plot_tree
plot_tree(xgb_final, feature_names=subset_train.columns, num_trees=0)
fig = matplotlib.pyplot.gcf()
fig.set_size_inches(150, 250)
fig.savefig('imgs/tree.png')

	id	listing_url	scrape_id	last_scraped	name	description	neighborhood_overview	picture_url	host_id	host_url	host_name	host_since	host_location	host_about	host_response_time	host_response_rate	host_acceptance_rate	host_is_superhost	host_thumbnail_url	host_picture_url	host_neighbourhood	host_listings_count	host_total_listings_count	host_verifications	host_has_profile_pic	host_identity_verified	neighbourhood	neighbourhood_cleansed	neighbourhood_group_cleansed	latitude	longitude	property_type	room_type	accommodates	bathrooms	bathrooms_text	bedrooms	beds	amenities	price	minimum_nights	maximum_nights	minimum_minimum_nights	maximum_minimum_nights	minimum_maximum_nights	maximum_maximum_nights	minimum_nights_avg_ntm	maximum_nights_avg_ntm	calendar_updated	has_availability	availability_30	availability_60	availability_90	availability_365	calendar_last_scraped	number_of_reviews	number_of_reviews_ltm	number_of_reviews_l30d	first_review	last_review	review_scores_rating	review_scores_accuracy	review_scores_cleanliness	review_scores_checkin	review_scores_communication	review_scores_location	review_scores_value	license	instant_bookable	calculated_host_listings_count	calculated_host_listings_count_entire_homes	calculated_host_listings_count_private_rooms	reviews_per_month
0	5456	https://www.airbnb.com/rooms/5456	20220312074014	2022-03-13	Walk to 6th, Rainey St and Convention Ctr	Great central location for walking to Convent...	My neighborhood is ideally located if you want...	https://a0.muscache.com/pictures/14084884/b5a3...	8028	https://www.airbnb.com/users/show/8028	Sylvia	2009-02-16	Austin, Texas, United States	I am a licensed Real Estate Broker and owner o...	within an hour	100%	94%	t	https://a0.muscache.com/im/users/8028/profile_...	https://a0.muscache.com/im/users/8028/profile_...	East Downtown	1.0	1.0	['email', 'phone', 'reviews', 'kba']	t	t	Austin, Texas, United States	78702	NaN	30.26057	-97.73441	Entire guesthouse	Entire home/apt	3	NaN	1 bath	1.0	2.0	["Patio or balcony", "Hair dryer", "Security c...	$240.00	2	90	2.0	3.0	90.0	90.0	2.0	90.0	NaN	t	7	18	44	305	2022-03-13	585	42	5	2009-03-08	2022-03-07	4.83	4.87	4.85	4.89	4.81	4.73	4.78	NaN	f	1	1	0	3.69
1	5769	https://www.airbnb.com/rooms/5769	20220312074014	2022-03-31	NW Austin Room	<b>The space</b><br />Looking for a comfortabl...	Quiet neighborhood with lots of trees and good...	https://a0.muscache.com/pictures/23822033/ac94...	8186	https://www.airbnb.com/users/show/8186	Elizabeth	2009-02-19	Austin, Texas, United States	We're easygoing professionals that enjoy meeti...	NaN	NaN	100%	t	https://a0.muscache.com/im/users/8186/profile_...	https://a0.muscache.com/im/users/8186/profile_...	SW Williamson Co.	1.0	1.0	['email', 'phone', 'reviews', 'jumio', 'govern...	t	t	Austin, Texas, United States	78729	NaN	30.45697	-97.78422	Private room in home	Private room	2	NaN	1 shared bath	1.0	1.0	["Essentials", "Hair dryer", "Hot water", "Hea...	$42.00	1	14	1.0	1.0	14.0	14.0	1.0	14.0	NaN	t	0	0	0	0	2022-03-31	264	3	0	2010-04-10	2021-07-03	4.90	4.90	4.86	4.91	4.94	4.74	4.92	NaN	f	1	0	1	1.81
2	6413	https://www.airbnb.com/rooms/6413	20220312074014	2022-03-31	Gem of a Studio near Downtown	Great studio apartment, perfect a single perso...	Travis Heights is one of the oldest neighborho...	https://a0.muscache.com/pictures/miso/Hosting-...	13879	https://www.airbnb.com/users/show/13879	Todd	2009-04-17	Austin, Texas, United States	We're a young family that likes to travel, we ...	within an hour	100%	100%	t	https://a0.muscache.com/im/pictures/user/4f35e...	https://a0.muscache.com/im/pictures/user/4f35e...	Travis Heights	1.0	1.0	['email', 'phone', 'reviews', 'jumio', 'offlin...	t	t	Austin, Texas, United States	78704	NaN	30.24885	-97.73587	Entire guesthouse	Entire home/apt	2	NaN	1 bath	NaN	1.0	["Kitchen", "Microwave", "Stove", "Shampoo", "...	$109.00	3	365	3.0	3.0	1125.0	1125.0	3.0	1125.0	NaN	t	0	0	0	0	2022-03-31	117	2	0	2009-12-14	2021-07-01	4.97	4.99	4.99	4.99	4.98	4.86	4.94	NaN	t	1	1	0	0.78
3	6448	https://www.airbnb.com/rooms/6448	20220312074014	2022-03-12	Secluded Studio @ Zilker - King Bed, Bright & ...	Clean, private space with everything you need ...	The neighborhood is fun and funky (but quiet)!...	https://a0.muscache.com/pictures/4513152/4ffc1...	14156	https://www.airbnb.com/users/show/14156	Amy	2009-04-20	Austin, Texas, United States	We are a family of four (with teenagers, all o...	within an hour	100%	98%	t	https://a0.muscache.com/im/users/14156/profile...	https://a0.muscache.com/im/users/14156/profile...	Zilker	2.0	2.0	['email', 'phone', 'facebook', 'reviews', 'kba']	t	t	Austin, Texas, United States	78704	NaN	30.26034	-97.76487	Entire guesthouse	Entire home/apt	3	NaN	1 bath	1.0	2.0	["Toaster", "Ceiling fan", "Outlet covers", "C...	$325.00	3	365	3.0	3.0	1125.0	1125.0	3.0	1125.0	NaN	t	9	19	39	107	2022-03-12	268	30	3	2011-09-06	2022-03-06	4.97	4.97	4.96	4.99	4.97	4.97	4.89	NaN	t	1	1	0	2.09
4	8502	https://www.airbnb.com/rooms/8502	20220312074014	2022-03-13	Woodland Studio Lodging	<b>The space</b><br />Fully furnished suite wi...	NaN	https://a0.muscache.com/pictures/miso/Hosting-...	25298	https://www.airbnb.com/users/show/25298	Karen	2009-07-11	Austin, Texas, United States	I handle the reservations at the studio on the...	within a day	83%	29%	f	https://a0.muscache.com/im/users/25298/profile...	https://a0.muscache.com/im/users/25298/profile...	East Riverside	1.0	1.0	['email', 'phone', 'reviews']	t	f	NaN	78741	NaN	30.23466	-97.73682	Entire guest suite	Entire home/apt	2	NaN	1 bath	1.0	1.0	["Essentials", "Kitchen", "Host greets you", "...	$44.00	3	90	3.0	4.0	90.0	90.0	3.4	90.0	NaN	t	0	0	26	26	2022-03-13	48	2	0	2010-02-19	2021-11-13	4.53	4.49	4.67	4.83	4.86	4.64	4.57	NaN	f	1	1	0	0.33

	host_tenure	host_listings_count	accommodates	minimum_nights	maximum_nights	number_of_reviews	number_of_reviews_l30d	host_response_time_unknown	host_response_time_within_a_day	host_response_time_within_a_few_hours	host_response_time_within_an_hour	host_response_rate_90-99	host_response_rate_Less_than_90	host_response_rate_unknown	host_acceptance_rate_70-80	host_acceptance_rate_81-90	host_acceptance_rate_91-99	host_acceptance_rate_Less_than_70	host_acceptance_rate_unknown	host_is_superhost_t	host_identity_verified_t	zipcode_78702	zipcode_78703	zipcode_78704	zipcode_78705	zipcode_78712	zipcode_78717	zipcode_78719	zipcode_78721	zipcode_78722	zipcode_78723	zipcode_78724	zipcode_78725	zipcode_78726	zipcode_78727	zipcode_78728	zipcode_78729	...	zipcode_78746	zipcode_78747	zipcode_78748	zipcode_78749	zipcode_78750	zipcode_78751	zipcode_78752	zipcode_78753	zipcode_78754	zipcode_78756	zipcode_78757	zipcode_78758	zipcode_78759	property_type_Hotel	property_type_House	property_type_Loft	property_type_Other	property_type_Rental_Unit	room_type_Private_room	room_type_Shared_room	has_availability_t	review_scores_rating_4.6-4.8	review_scores_rating_4.8-4.99	review_scores_rating_5	review_scores_rating_Less_than_4	review_scores_rating_unknown	instant_bookable_t	in_texas_Yes	in_texas_unknown	pool_True	kitchen_True	free_parking_True	free_street_parking_True	hot_tub_True	washer_True	workspace_True	gym_True
8055	725	0.0	1	90	100	0	0	0	1	0	0	1	0	0	0	1	0	0	0	1	1	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0	0	0	1	0	1	0	0	0	0	1	0	1	0	1	1	1	0	0	1	1	0
2570	2976	1.0	1	1	1125	3	0	1	0	0	0	0	0	1	0	0	0	0	1	0	1	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0	0	0	1	0	0	0	0	0	0	0	0	0	0	0	1	1	0	0	1	0	0
8396	822	9.0	8	1	112	4	0	0	0	0	1	0	0	0	0	0	1	0	0	0	1	0	0	0	0	0	0	0	1	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0	0	0	0	0	1	1	0	0	0	0	1	0	0	0	1	1	1	1	1	1	0
618	2945	1.0	2	1	1125	1	0	1	0	0	0	0	0	1	0	0	0	0	1	0	1	0	0	0	1	0	0	0	0	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0	0	0	0	1	1	0	0	0	1	0	0	0	0	1	1	0	0	0	1	0	0
7983	3683	5.0	6	1	1125	74	2	0	0	0	1	0	0	0	0	0	0	0	0	1	1	0	0	0	1	0	0	0	0	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0	0	0	0	0	1	0	1	0	0	0	1	1	0	0	1	0	1	0	1	0	0
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
5816	932	0.0	2	1	30	6	0	1	0	0	0	0	0	1	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	1	0	0	0	0	0	0	0	1	1	0	0	0	0	1	0	0	0	1	0	1	1	1	0	0	1	0	0
9559	3336	27.0	4	2	1125	14	1	0	0	0	1	1	0	0	0	0	0	0	0	0	1	0	0	0	1	0	0	0	0	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0	0	0	0	0	1	1	0	0	1	0	0	0	1	1	0
9798	1014	0.0	6	30	365	0	0	0	0	0	1	0	0	0	0	0	1	0	0	1	0	0	0	0	0	0	0	0	1	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0	0	0	0	0	1	0	0	0	0	1	1	1	0	0	1	1	0	0	1	1	0
11276	2484	3.0	7	1	365	5	5	0	0	0	1	0	0	0	0	0	0	0	0	0	1	0	0	0	0	0	0	0	1	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0	0	0	0	0	1	0	0	1	0	0	1	1	0	0	1	1	1	0	1	1	0
1774	3100	2.0	8	2	1125	0	0	1	0	0	0	0	0	1	0	0	0	0	1	0	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	1	0	0	0	0	1	0	0	0	0	0	1	0	0	0	0	1	0	0	0	0	1	1	0	0	1	1	0

	Column	Count	%
0	neighbourhood	4913	41.037421
1	host_response_time	3449	28.808887
2	host_response_rate	3449	28.808887
3	review_scores_value	3020	25.225526
4	review_scores_location	3020	25.225526
5	review_scores_checkin	3019	25.217173
6	review_scores_accuracy	3018	25.208821
7	review_scores_communication	3018	25.208821
8	review_scores_cleanliness	3018	25.208821
9	first_review	2946	24.607417
10	last_review	2946	24.607417
11	review_scores_rating	2946	24.607417
12	reviews_per_month	2946	24.607417
13	host_acceptance_rate	2862	23.905780
14	host_neighbourhood	1718	14.350150
15	bedrooms	711	5.938857
16	beds	150	1.252923
17	host_location	18	0.150351
18	bathrooms_text	16	0.133645
19	host_since	3	0.025058
20	host_identity_verified	3	0.025058
21	host_has_profile_pic	3	0.025058
22	host_total_listings_count	3	0.025058
23	host_listings_count	3	0.025058
24	host_is_superhost	3	0.025058
25	minimum_maximum_nights	1	0.008353
26	maximum_maximum_nights	1	0.008353
27	maximum_nights_avg_ntm	1	0.008353
28	minimum_nights_avg_ntm	1	0.008353
29	maximum_minimum_nights	1	0.008353
30	minimum_minimum_nights	1	0.008353
31	longitude	0	0.000000
32	calculated_host_listings_count	0	0.000000
33	instant_bookable	0	0.000000
34	calculated_host_listings_count_entire_homes	0	0.000000
35	calculated_host_listings_count_private_rooms	0	0.000000
36	calculated_host_listings_count_shared_rooms	0	0.000000
37	neighbourhood_cleansed	0	0.000000
38	latitude	0	0.000000
39	accommodates	0	0.000000
40	property_type	0	0.000000
41	number_of_reviews_l30d	0	0.000000
42	amenities	0	0.000000
43	number_of_reviews	0	0.000000
44	availability_365	0	0.000000
45	availability_90	0	0.000000
46	availability_60	0	0.000000
47	availability_30	0	0.000000
48	has_availability	0	0.000000
49	room_type	0	0.000000
50	maximum_nights	0	0.000000
51	minimum_nights	0	0.000000
52	price	0	0.000000
53	number_of_reviews_ltm	0	0.000000

	host_since	host_location	host_response_time	host_response_rate	host_acceptance_rate	host_is_superhost	host_neighbourhood	host_listings_count	host_total_listings_count	host_has_profile_pic	host_identity_verified	neighbourhood	neighbourhood_cleansed	latitude	longitude	property_type	room_type	accommodates	bathrooms_text	bedrooms	beds	amenities	price	minimum_nights	maximum_nights	minimum_minimum_nights	maximum_minimum_nights	minimum_maximum_nights	maximum_maximum_nights	minimum_nights_avg_ntm	maximum_nights_avg_ntm	has_availability	number_of_reviews	first_review	last_review	review_scores_rating	instant_bookable	calculated_host_listings_count	calculated_host_listings_count_entire_homes	calculated_host_listings_count_private_rooms	reviews_per_month
881	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	78704	30.25492	-97.74532	Entire rental unit	Entire home/apt	2	1 bath	1.0	1.0	["Kitchen", "Smoke alarm", "Long term stays al...	$75.00	1	1125	1.0	1.0	1125.0	1125.0	1.0	1125.0	f	1	2015-08-13	2015-08-13	0.00	f	1	1	0	0.01
807	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	78704	30.25261	-97.74740	Entire rental unit	Entire home/apt	4	1 bath	2.0	3.0	["Free parking on premises", "Essentials", "Sm...	$400.00	2	3	2.0	2.0	3.0	3.0	2.0	3.0	f	0	NaN	NaN	NaN	f	1	1	0	NaN
1491	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	78741	30.22839	-97.72529	Private room in rental unit	Private room	1	1 bath	1.0	1.0	["Essentials", "Heating", "Smoke alarm", "Kitc...	$27.00	1	5	1.0	1.0	5.0	5.0	1.0	5.0	t	4	2015-09-07	2015-09-24	4.67	f	1	0	1	0.05

	host_tenure	host_response_time	host_response_rate	host_acceptance_rate	host_is_superhost	host_listings_count	host_has_profile_pic	host_identity_verified	zipcode	latitude	longitude	property_type	room_type	accommodates	bedrooms	beds	price	minimum_nights	maximum_nights	minimum_minimum_nights	maximum_minimum_nights	minimum_maximum_nights	maximum_maximum_nights	minimum_nights_avg_ntm	maximum_nights_avg_ntm	has_availability	availability_30	availability_60	availability_90	availability_365	number_of_reviews	number_of_reviews_ltm	first_review	last_review	review_scores_rating	instant_bookable	reviews_per_month	in_texas	bathrooms	pool	wifi	kitchen	free parking	free street parking	hot tub	washer	air conditioning	workspace	pet	gym
5022	3341	within a few hours	100	81-90	t	1.0	t	t	78702	30.27573	-97.70442	House	Entire home/apt	2	NaN	1.0	$99.00	3	1125	3.0	3.0	1125.0	1125.0	3.0	1125.0	t	1	11	41	131	11	4	2019-03-15	2021-10-26	4.91	f	0.30	Yes	1.0	False	True	False	False	True	False	False	True	False	False	False
1067	2684	within a few hours	100	100	f	0.0	t	t	78722	30.28148	-97.71742	House	Private room	1	1.0	1.0	$44.00	30	360	30.0	30.0	360.0	360.0	30.0	360.0	t	28	57	86	174	9	0	2015-03-19	2016-04-04	5.00	f	0.11	Yes	1.0	False	True	True	True	False	False	True	True	True	False	False
604	2948	unknown	unknown	unknown	f	1.0	t	f	78705	30.29925	-97.73819	House	Entire home/apt	10	4.0	6.0	$1,200.00	1	1125	1.0	1.0	1125.0	1125.0	1.0	1125.0	f	0	0	0	0	0	0	NaN	NaN	NaN	f	NaN	No	3.0	False	True	True	True	False	False	True	True	False	False	False
1750	2701	unknown	unknown	unknown	f	1.0	t	t	78751	30.30511	-97.72662	House	Entire home/apt	2	NaN	1.0	$75.00	1	1125	1.0	1.0	1125.0	1125.0	1.0	1125.0	f	0	0	0	0	0	0	NaN	NaN	NaN	f	NaN	Yes	1.0	False	False	True	False	False	False	False	True	False	False	False
8342	342	within a day	Less than 90	81-90	f	5.0	t	t	78704	30.23769	-97.77817	House	Entire home/apt	6	3.0	3.0	$439.00	2	1125	2.0	2.0	1125.0	1125.0	2.0	1125.0	t	0	0	0	0	13	13	2021-07-09	2021-11-15	4.15	f	1.47	No	2.5	False	True	True	False	False	False	True	True	True	False	False

	host_tenure	host_response_time	host_response_rate	host_acceptance_rate	host_is_superhost	host_listings_count	host_has_profile_pic	host_identity_verified	zipcode	latitude	longitude	property_type	room_type	accommodates	bedrooms	beds	price	minimum_nights	maximum_nights	minimum_minimum_nights	maximum_minimum_nights	minimum_maximum_nights	maximum_maximum_nights	minimum_nights_avg_ntm	maximum_nights_avg_ntm	has_availability	availability_30	availability_60	availability_90	availability_365	number_of_reviews	first_review	last_review	review_scores_rating	instant_bookable	reviews_per_month	in_texas	bathrooms	pool	kitchen	free parking	free street parking	hot tub	washer	workspace	gym
6507	787	a few days or more	unknown	unknown	f	0.0	t	t	78702	30.26065	-97.72315	Hotel	Private room	2	NaN	1.0	10000.0	1	1125	1.0	1.0	1125.0	1125.0	1.0	1125.0	t	30	60	90	365	0	NaN	NaN	NaN	t	NaN	No	1.0	False	False	False	False	False	False	False	False
10559	725	within a day	90-99	81-90	t	0.0	t	t	78702	30.25627	-97.73541	House	Entire home/apt	16	13.0	23.0	20000.0	3	7	3.0	3.0	7.0	7.0	3.0	7.0	t	7	7	7	260	0	NaN	NaN	NaN	f	NaN	Yes	7.5	True	True	True	False	False	True	True	False
3713	4057	unknown	unknown	unknown	f	1.0	t	t	78701	30.26809	-97.74971	Rental Unit	Entire home/apt	5	1.0	3.0	10000.0	500	1125	500.0	500.0	1125.0	1125.0	500.0	1125.0	t	30	60	90	180	1	2018-03-14	2018-03-14	5.0	f	0.02	Yes	1.0	True	True	False	False	False	True	True	True
10663	877	a few days or more	unknown	70-80	f	0.0	t	t	78702	30.26370	-97.73168	Hotel	Private room	4	1.0	1.0	10000.0	1	365	1.0	1.0	365.0	365.0	1.0	365.0	t	30	60	90	364	0	NaN	NaN	NaN	t	NaN	Yes	1.0	True	False	False	False	False	False	True	False
7312	1308	within an hour	100	Less than 70	f	46.0	t	t	78733	30.34310	-97.87865	House	Entire home/apt	16	8.0	15.0	14795.0	1	365	1.0	3.0	2.0	1125.0	2.3	247.6	t	28	58	88	360	0	NaN	NaN	NaN	t	NaN	Yes	7.5	True	True	True	False	False	True	True	True

	Predictor	Variance Inflation Factor
13	host_response_rate_unknown	64.190319
10	host_response_time_within_an_hour	43.640061
7	host_response_time_unknown	43.356595
81	kitchen_True	21.188898
71	has_availability_t	14.065250
85	washer_True	10.110810
65	property_type_House	8.529728
0	host_tenure	7.440383
9	host_response_time_within_a_few_hours	7.013602
18	host_acceptance_rate_unknown	6.783710
20	host_identity_verified_t	6.779704
82	free_parking_True	6.003626
78	in_texas_Yes	5.756088
2	accommodates	5.245320
73	review_scores_rating_4.8-4.99	5.096264

	feature	coef	magnitude
87	accommodates	0.385814	0.385814
0	room_type_Private_room	-0.227003	0.227003
86	property_type_House	0.116002	0.116002
1	room_type_Shared_room	-0.104623	0.104623
85	review_scores_rating_unknown	0.094790	0.094790
84	property_type_Hotel	0.067476	0.067476

	Model	R^2	MSE
0	Linear Regression	0.527486	0.393278
1	Ridge Regression	0.562832	0.363859
2	XGB Regressor	0.637181	0.301977

Project Introduction¶

Data Description¶

Data Cleaning¶

Initial Look¶

Cleaning predictors, Handling Missingness and Univariate Analysis¶

host_since¶

host_location¶

host_response_time¶

host_response_rate¶

host_acceptance_rate¶

host_neighbourhood¶

host_listings_count¶

neighbourhood, neighbourhood_cleansed¶

Property Type and Room Type¶

bathrooms_text¶

amenities¶

price¶

minimum_nights and maximum_nights¶

has_availability¶

number_of_reviews_ltm¶

first_review and last_review¶

review_scores_rating¶

beds and bedrooms¶

Post-Cleaning Thoughts¶

Exploratory Data Analysis¶

host_tenure¶

host_response_time¶

host_is_superhost¶

host_has_profile_pic¶

property_type and room_type¶

accomodates, bedrooms, beds¶

Latitude, Longitude and zipcode¶

Model Preparation¶

Testing Multicollinearity¶

Modeling¶

Model 1: Log Normal Linear Regression¶

Linear Regression Assumptions¶

Sequential Forward Selection¶

Ridge Regression¶

Extreme Gradient Boosted Decision Trees (XGBoost)¶

Model Decision and Final Tuning¶

Fine tuning Parameters¶

Fitting with final model parameters¶

Feature Importance¶

Feature Reduction¶

Final Model after Feature Reduction¶

Testing our final model¶

Conclusion¶

Model Summary¶

What Didn't Work?¶

What about the Unexplained Variance?¶

Future Reccommendations¶

Further Review¶

`host_since`¶

`host_location`¶

`host_response_time`¶

`host_response_rate`¶

`host_acceptance_rate`¶

`host_neighbourhood`¶

`host_listings_count`¶

`neighbourhood`, `neighbourhood_cleansed`¶

`Property Type` and `Room Type`¶

`bathrooms_text`¶

`amenities`¶

`price`¶

`minimum_nights` and `maximum_nights`¶

`first_review` and `last_review`¶

`beds` and `bedrooms`¶

`host_tenure`¶

`host_response_time`¶

`host_is_superhost`¶

`host_has_profile_pic`¶

`property_type` and `room_type`¶

`accomodates`, `bedrooms`, `beds`¶

`Latitude`, `Longitude` and `zipcode`¶